In such a spread-spectrum communication system, a carrier is modulated by a pseudo-noise code (hereinafter called PN code) which is a binary code and also is modulated by data, as shown in FIG. 1(A). In FIG. 1A, reference numeral 1 designates a data source, 2 is a modulator, 3 is a PN code generator, 4 is a carrier wave generator, 5 is a modulator, and 6 is a transmitter. At a receiver's end, as shown in FIG. 1(B), a matched filter detects a correlation between the arriving signal and a reference PN code therein so that when both the codes coincide or are slightly displaced, an autocorrelation waveform (hereinafter called "correlation spike waveform") with a relatively large amplitude is processed and demodulated into the data. In FIG. 1B, reference numeral 7 refers to a receiving antenna, 8 is a correlator, 9 is a reference PN code generator, 10 is a data demodulator, and 11 is a data stream.
A convolver is used as a matched filter. A convolver in general is a functional unit to effect a convolution integral, but may be a matched filter to effect a correlation operation if a binary reference code (hereinafter called "reference code") is the time-reversed image of the received code.
A surface acoustic wave convolver (hereinafter called "SAW convolver") is referred to as one type of convolver. The industry provides different structures of SAW convolvers, i.e., a lamination of a piezoelectric substance and silicon spaced by a gap, a combination of a piezoelectric substance and silicon united via an oxide layer, a single body of a piezoelectric substance, etc. They all have nonlinear properties and perform a correlation operation of two signals by use of an interaction therebetween, the result of which is integrated by a gate electrode provided in an interaction region.
FIG. 2 shows a construction of a SAW convolver. Reference numerals 12 and 13 are transducers, 14 is a piezoelectric substance, 15 is an oxide layer, 16 is a silicon substrate, and 17 is a gate electrode. A signal s(t) entered from the transducer 12 travels to the right in the Figure, and a signal r(t) entered from the transducer 13 travels to the left. Since the construction comprising the piezoelectric layer, oxide layer and silicon has a nonlinear property, an interaction occurs between the signals s(t) and r(t) to cause the correlation operation and integrate the result therefrom in the gate electrode 17.
A signal c(t) produced by the gate electrode 17 is expressed by: ##EQU1## Where A is a constant, T is the time required for an acoustic wave to pass under the gate electrode (hereinafter called "under-gate delay time"), x is the propagation distance of the signal s(t), and s is the sonic speed.
A PN code in general has a given cycle. In formation of a waveform at the transmitter's end, it is often employed to correlate one cycle of the PN code with one bit length of the data. For a better explanation or understanding, the instant description takes an example where one cycle of the PN code equals the length of one data bit.
The relationship between the under-gate delay time and the PN code may be selected as desired. Namely, the under-gate delay time may be shorter, equal or longer with respect to one cycle of the PN code. The under-gate delay time means the integral period in the correlation operation. The integral period preferably equals one cycle of the PN code, considering the correlation characteristics of the PN code. The instant description takes an example where the under-gate delay time equals one cycle of the PN code.
The above-cited relationships are shown FIGS. 3A, 3B, and 3C. FIG. 3A shows a data bit, and FIG. 3B shows an arrangement of the PN code. The FIGS. 3A and 3B show that the length l of one data bit equals one cycle of the PN code. FIG. 3C is a diagrammatic cross-sectional view of a convolver wherein the delay time in the length L of the gate electrode equals l. Here again, the illustrated arrangement is simply an example, and any relationship may be selected between one data bit, one cycle of the PN code and the under-gate delay time.
In practical communication, the receiver always stands by for reception, with the reference signal entered in one of the transducers. When a signal is received, it is supplied from the other transducer to the convolver. If the PN codes involved in the received signal coincides with the reference signal, the gate electrode of the convolver provides a correlation spike waveform. However, it is still unknown in which position the both codes are aligned. The data is not demodulated into its proper form unless the alignment is established at a proper position. For example, if both the codes are aligned at the position shown by FIG. 4A, the received PN code is shared half and half by data bits A and B. In FIGS. 4A and 4B, D shows a data bit, R shows the received PN code, RP shows the reference PN code, and L shows the region under the gate electrode where the interaction occurs. Ais the time-reversed PN code of the PN code A.
As described above, some means is necessary to finally make both the codes coincide at the position of FIG. 4B if they first coincide in any other position. The first coincidence of both the codes after reception of the arriving signal and until the coincidence at the position of FIG. 4B is called "primary synchronization" in this text. Although the primary synchronization is out of the scope of the present invention, it is described in a report by D. Brodtkorb and J. E. Laynor entitled "Fast synchronization in a spread-spectrum system based on acoustoelectric convolvers" and printed in pages 561 through 566 of 1978 Ultrasonics Symposium Proceedings, IEEE Cat. No. 78CH1344-ISU.
After the primary synchronization is established, and the placement of FIG. 4B is once obtained, a possible difference between the clock frequency of the received PN code and the clock frequency of the reference PN code gradually displaces the alignment position from the placement of FIG. 4B. The displacement at every instant of encounter of the heads of the received and reference PN codes is expressed by: ##EQU2## where .function.r is the clock frequency of the reference PN code, .function.t is the clock frequency of the received PN code, and N is the number of chips in one cycle of the PN code.
Regardless of the primary synchronization, a possible difference between the clock frequencies of both the codes gradually displaces the alignment position from the proper placement and disables demodulation of the data. This means that clock oscillators with an exactly uniform clock frequency must be provided in the transmitter and the receiver. Such clock oscillators are normally based on quartz oscillators. However, it is extremely difficult to manufacture a plurality of quartz oscillators with an exactly uniform frequency. Besides this, they require a strict control of the environment such as temperature and humidity.
The report by D. Brodtkorb et al. discloses how to overcome the problem. They use a counter to detect the displacement and shift time base of the reference PN code so as to cure the displacement. The misalignment correction and subsequent maintenance of the perfect alignment at the proper position are called "synchronization maintenance" in this text. The proposal by D. Brodtkorb et al., however, has a drawback in that the signal processing is complicated because it requires operation to drive or stop the counter.